Advanced Machine Learning Models for Estimating the Distribution of Sea-Surface Particulate Organic Carbon (POC) Concentrations Using Satellite Remote Sensing Data: The Mediterranean as an Example
Abstract
:1. Introduction
2. Materials and Methods
2.1. Study Area
2.2. Data Sources
2.2.1. In Situ Data
2.2.2. Satellite Data
2.2.3. Agreement between In Situ and Satellite Data
2.3. Feature Selection
2.4. Model Selection
2.5. Dual Optimization of the RF Model
2.5.1. Initial Optimization of the RF Model
2.5.2. Re-Optimization of the RF Model
2.6. Statistical Indicators Used for Model Development, Validation, and Testing
3. Results and Discussion
3.1. Feature Selection
3.2. Machine Learning Models for POC
3.2.1. Accuracy of the Models for Different Datasets
3.2.2. Re-Optimization of the RF Model
3.3. Assessment of the TRRF Model
3.3.1. Comparison of TRRF Products and NASA Products
3.3.2. Comparison between TRRF Products, NASA Products, and Actual Measured Values
3.3.3. The Impact of Human Activities on the Distribution of Surface POC in the Mediterranean Sea
4. Conclusions
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Acknowledgments
Conflicts of Interest
References
- Stramski, D.; Reynolds, R.A.; Kahru, M.; Mitchell, B.G. Estimation of particulate organic carbon in the ocean from satellite remote sensing. Science 1999, 285, 239–242. [Google Scholar] [CrossRef]
- Woźniak, S.B.; Darecki, M.; Zabłocka, M.; Burska, D.; Dera, J. New simple statistical formulas for estimating surface concentrations of suspended particulate matter (SPM) and particulate organic carbon (POC) from remote-sensing reflectance in the southern Baltic Sea. Oceanologia 2016, 58, 161–175. [Google Scholar] [CrossRef]
- Son, Y.B.; Gardner, W.D.; Mishonov, A.V.; Richardson, M.J. Multispectral remote-sensing algorithms for particulate organic carbon (POC): The Gulf of Mexico. Remote Sens. Env. 2009, 113, 50–61. [Google Scholar] [CrossRef]
- Stramski, D.; Reynolds, R.A.; Babin, M.; Kaczmarek, S.; Lewis, M.R.; Röttgers, R.; Sciandra, A.; Stramska, M.; Twardowski, M.S.; Franz, B.A. Relationships between the surface concentration of particulate organic carbon and optical properties in the eastern South Pacific and eastern Atlantic Oceans. Biogeosciences 2008, 5, 171–201. [Google Scholar] [CrossRef]
- Le, C.; Lehrter, J.C.; Hu, C.; MacIntyre, H.; Beck, M.W. Satellite observation of particulate organic carbon dynamics on the L ouisiana continental shelf. J. Geophys. Res. Ocean. 2017, 122, 555–569. [Google Scholar] [CrossRef]
- Stramski, D.; Joshi, I.; Reynolds, R.A. Ocean color algorithms to estimate the concentration of particulate organic carbon in surface waters of the global ocean in support of a long-term data record from multiple satellite missions. Remote Sens. Environ. 2022, 269, 112776. [Google Scholar] [CrossRef]
- Cai, S.; Wu, M.; Le, C. Satellite observation of the long-term dynamics of particulate organic carbon in the East China Sea based on a hybrid algorithm. Remote Sens. 2022, 14, 3220. [Google Scholar] [CrossRef]
- Allison, D.B.; Stramski, D.; Mitchell, B.G. Empirical ocean color algorithms for estimating particulate organic carbon in the Southern Ocean. J. Geophys. Res. Ocean. 2010, 115. [Google Scholar] [CrossRef]
- Sauzède, R.; Johnson, J.E.; Claustre, H.; Camps-Valls, G.; Ruescas, A.B. Estimation of oceanic particulate organic carbon with machine learning. Isprs Ann. Photogramm. Remote Sens. Spat. Inf. Sci. 2020, 2, 949–956. [Google Scholar] [CrossRef]
- Fellous, S.; Bendjama, A.; Benzaoui, Y. Use of machine learning algorithms and in situ data for estimating particulate organic carbon from the Mediterranean Sea. Larhyss J. 2023, 56, 179–192, P-Issn 1112-3680/E-Issn 2521-9782. [Google Scholar]
- Wu, J.; Chen, X.; Zhang, H.; Xiong, L.; Lei, H.; Deng, S. Hyperparameter optimization for machine learning models based on Bayesian optimization. J. Electron. Sci. Technol. 2019, 17, 26–40. [Google Scholar]
- Alibrahim, H.; Ludwig, S.A. Hyperparameter optimization: Comparing genetic algorithm against grid search and bayesian optimization. In Proceedings of the 2021 IEEE Congress on Evolutionary Computation (CEC), Kraków, Poland, 28 June–1 July 2021; pp. 1551–1559. [Google Scholar]
- Probst, P.; Wright, M.N.; Boulesteix, A.L. Hyperparameters and tuning strategies for random forest. Wiley Interdiscip. Rev. Data Min. Knowl. Discov. 2019, 9, e1301. [Google Scholar] [CrossRef]
- Wang, J.; Zhang, T.; Fu, B. A measure of spatial stratified heterogeneity. Ecol. Indic. 2016, 67, 250–256. [Google Scholar] [CrossRef]
- Tanhua, T.; Hainbucher, D.; Schroeder, K.; Cardin, V.; Álvarez, M.; Civitarese, G. The Mediterranean Sea system: A review and an introduction to the special issue. Ocean. Sci. 2013, 9, 789–803. [Google Scholar] [CrossRef]
- Stramska, M.; Stramski, D. Variability of particulate organic carbon concentration in the north polar Atlantic based on ocean color observations with Sea-viewing Wide Field-of-view Sensor (SeaWiFS). J. Geophys. Res. Ocean. 2005, 110. [Google Scholar] [CrossRef]
- Werdell, P.J.; Fargion, G.S.; McClain, C.R.; Bailey, S.W. The SeaWiFS Bio-Optical Archive and Storage System (Sea-BASS): Current Architecture and Implementation; NASA: Washington, DC, USA, 2002. [Google Scholar]
- Song, Y.; Wang, J.; Ge, Y.; Xu, C. An optimal parameters-based geographical detector model enhances geographic characteristics of explanatory variables for spatial heterogeneity analysis: Cases with different types of spatial data. GIScience Remote Sens. 2020, 57, 593–610. [Google Scholar] [CrossRef]
- Wang, X.; Meng, Q.; Zhang, L.; Hu, D. Evaluation of urban green space in terms of thermal environmental benefits using geographical detector analysis. Int. J. Appl. Earth Obs. 2021, 105, 102610. [Google Scholar] [CrossRef]
- Zhang, Z.; Song, Y.; Wu, P. Robust geographical detector. Int. J. Appl. Earth Obs. 2022, 109, 102782. [Google Scholar] [CrossRef]
- Wen, Z.; Wang, Q.; Ma, Y.; Jacinthe, P.A.; Liu, G.; Li, S.; Shang, Y.; Tao, H.; Fang, C.; Lyu, L. Remote estimates of suspended particulate matter in global lakes using machine learning models. Int. Soil Water Conserv. Res. 2024, 12, 200–216. [Google Scholar] [CrossRef]
- Liu, H.; Li, Q.; Bai, Y.; Yang, C.; Wang, J.; Zhou, Q.; Hu, S.; Shi, T.; Liao, X.; Wu, G. Improving satellite retrieval of oceanic particulate organic carbon concentrations using machine learning methods. Remote Sens. Environ. 2021, 256, 112316. [Google Scholar] [CrossRef]
- Breiman, L. Random forests. Mach. Learn. 2001, 45, 5–32. [Google Scholar] [CrossRef]
- Belgiu, M.; Drăguţ, L. Random forest in remote sensing: A review of applications and future directions. ISPRS J. Photogramm. 2016, 114, 24–31. [Google Scholar] [CrossRef]
- Hultquist, C.; Chen, G.; Zhao, K. A comparison of Gaussian process regression, random forests and support vector regression for burn severity assessment in diseased forests. Remote Sens. Lett. 2014, 5, 723–732. [Google Scholar] [CrossRef]
- Wang, S.; Jin, X.; Adhikari, K.; Li, W.; Yu, M.; Bian, Z.; Wang, Q. Mapping total soil nitrogen from a site in north-eastern China. Catena 2018, 166, 134–146. [Google Scholar] [CrossRef]
- Ying, C.; Qi-Guang, M.; Jia-Chen, L.; Lin, G. Advance and prospects of AdaBoost algorithm. Acta Autom. Sin. 2013, 39, 745–758. [Google Scholar]
- McCulloch, W.S.; Pitts, W. A logical calculus of the ideas immanent in nervous activity. Bull. Math. Biophys. 1943, 5, 115–133. [Google Scholar] [CrossRef]
- Brochu, E.; Cora, V.M.; De Freitas, N. A tutorial on Bayesian optimization of expensive cost functions, with application to active user modeling and hierarchical reinforcement learning. arXiv 2010, arXiv:1012.2599. [Google Scholar]
- Snoek, J.; Larochelle, H.; Adams, R.P. Practical bayesian optimization of machine learning algorithms. In Proceedings of the Advances in Neural Information Processing Systems 25, Lake Tahoe, NV, USA, 3–6 December 2012. [Google Scholar]
- Mockus, J.; Mockus, J. The Bayesian Approach to Local Optimization; Springer: Berlin/Heidelberg, Germany, 1989. [Google Scholar]
- Wang, X.; Jin, Y.; Schmitt, S.; Olhofer, M. Recent advances in Bayesian optimization. ACM Comput. Surv. 2023, 55, 1–36. [Google Scholar] [CrossRef]
- Chalupka, K.; Williams, C.K.; Murray, I. A Framework for Evaluating Approximation Methods for Gaussian Process Regression. J. Mach. Learn. Res. 2013, 14, 333–350. [Google Scholar]
- Jones, D.R.; Schonlau, M.; Welch, W.J. Efficient global optimization of expensive black-box functions. J. Global Optim. 1998, 13, 455–492. [Google Scholar] [CrossRef]
- Zaneveld, J.R.V. Light and water: Radiative transfer in natural waters. JSTOR 1995, 76, 60–63. [Google Scholar]
- Cavan, E.L.; Trimmer, M.; Shelley, F.; Sanders, R. Remineralization of particulate organic carbon in an ocean oxygen minimum zone. Nat. Commun. 2017, 8, 14847. [Google Scholar] [CrossRef]
- Stelzer, R.S.; Thad Scott, J.; Bartsch, L.A.; Parr, T.B. Particulate organic matter quality influences nitrate retention and denitrification in stream sediments: Evidence from a carbon burial experiment. Biogeochemistry 2014, 119, 387–402. [Google Scholar] [CrossRef]
- Egge, J.K.; Jacobsen, A. Influence of silicate on particulate carbon production in phytoplankton. Mar. Ecol. Prog. Ser. 1997, 147, 219–230. [Google Scholar] [CrossRef]
- Guo, M.; Li, X.; Wang, Y.; Zhang, Y.; Fu, Q.; Huguet, A.; Liu, G. New insights into the mechanism of phosphate re-lease during particulate organic matter photodegradation based on optical and molecular signatures. Water Res. 2023, 236, 119954. [Google Scholar] [CrossRef]
- Wu, H.; Cui, L.; Wang, L.; Sun, R.; Zheng, Z. A method for estimating particulate organic carbon at the sea surface based on geodetector and machine learning. Front. Mar. Sci. 2023, 10, 1295874. [Google Scholar] [CrossRef]
- Huang, C.; Lao, Q.; Chen, F.; Zhang, S.; Chen, C.; Bian, P.; Zhu, Q. Distribution and sources of particulate organic matter in the Northern South China Sea: Implications of human activity. J. Ocean. Univ. China 2021, 20, 1136–1146. [Google Scholar] [CrossRef]
- Ming, Y.; Gao, L.; Guo, L. Dissolved and particulate organic carbon dynamics in the lower Changjiang River on timescales from seasonal to decades: Response to climate and human impacts. J. Mar. Syst. 2023, 239, 103855. [Google Scholar] [CrossRef]
- Cori, B. Spatial dynamics of Mediterranean coastal regions. J. Coast. Conserv. 1999, 5, 105–112. [Google Scholar] [CrossRef]
- Lloret, J.; Riera, V. Evolution of a Mediterranean coastal zone: Human impacts on the marine environment of Cape Creus. Environ. Manag. 2008, 42, 977–988. [Google Scholar] [CrossRef]
- Maciejewska, A.; Pempkowiak, J. DOC and POC in the water column of the southern Baltic. Part I. Evaluation of factors influencing sources, distribution and concentration dynamics of organic matter. Oceanologia 2014, 56, 523–548. [Google Scholar] [CrossRef]
Parameter | Full Name | Transducers/Product ID | Spatial Resolution | Time Resolution | Amount |
---|---|---|---|---|---|
BBP | backscattering coefficient of particles | OCEANCOLOUR_GLO_BGC_L3_MY_009_103 | 4 × 4 km | daily | 37 |
CDM | volume absorption coefficient of radiative flux in seawater due to dissolved organic matter and non-algal particles | OCEANCOLOUR_GLO_B GC_L3_MY_009_103 | 4 × 4 km | daily | 37 |
Kd_490 | diffuse attenuation coefficient at 490 nm | OCEANCOLOUR_GLO_BGC_L3_MY_009_103 | 4 × 4 km | daily | 37 |
Rrs (412, 443, 490, 555, and 670) | remote sensing reflectance (412 nm, 443 nm, 490 nm, 555 nm, and 670 nm) | OCEANCOLOUR_GLO_BGC_L3_MY_009_103 | 4 × 4 km | daily | 37 |
Rrs (547, 645, 667, 469, 488, and 510) | remote sensing reflectance (547 nm, 645 nm, 490 nm, 488 nm, and 510 nm) | MODIS Aqua | 4 × 4 km | daily | 37 |
SPM | suspended particulate matter | OCEANCOLOUR_GLO_B GC_L3_MY_009_103 | 4 × 4 km | daily | 37 |
SSTemp | sea-surface temperature | METOFFICE-GLO-SST-L4-REP-OBS-SST | 0.05° × 0.05° | daily | 37 |
SSS | seawater salinity | GLOBAL_MULTIY EAR_PHY_001_030 | 0.083° × 0.083° | daily | 37 |
ZEU | euphotic zone depth | GLOBAL_MULTIYE AR_BGC_001_033 | 0.083° × 0.083° | daily | 37 |
Dos | sea-surface density | MUTOBS_GLO_PHY_S_SURFACE_MYNRT_015_013 | 0.125° × 0.125° | daily | 37 |
Chl | chlorophyll a | GLOBAL_MULTIY EAR_BGC_001_029 | 0.25° × 0.25° | daily | 37 |
Mld | ocean mixed-layer thickness | MULTIOBS_GLO_PHY_TSUV_3D_MYNRT_015_012 | 0.25° × 0.25° | weekly | 7 |
NO3 | nitrate | GLOBAL_MULTIY EAR_BGC_001_029 | 0.25° × 0.25° | daily | 37 |
O2 | oxygen | GLOBAL_MULTIY EAR_BGC_001_029 | 0.25° × 0.25° | daily | 37 |
pH | potential of hydrogen | cmems_mod_glo_bgc_my_0.25_P1M-m | 0.25° × 0.25° | weekly | 7 |
PO4 | phosphate | GLOBAL_MULTIY EAR_BGC_001_029 | 0.25° × 0.25° | daily | 37 |
SiO3 | silicate | GLOBAL_MULTIY EAR_BGC_001_029 | 0.25° × 0.25° | daily | 37 |
Ugos | geostrophic eastward ocean velocity | SEALEVEL_GLO_P HY_L4_MY_008_047 | 0.25° × 0.25° | daily | 37 |
Vgos | geostrophic northward ocean velocity | SEALEVEL_GLO_P HY_L4_MY_008_047 | 0.25° × 0.25° | daily | 37 |
Dataset | Bias | Variance | R2 | RMSE | MAPE | |
---|---|---|---|---|---|---|
Training | BPNN | 0.054 | 0.816 | 0.93 | 0.27 | 4.02% |
XGBoost | −3.040 | 0.005 | 0.99 | 0.01 | 0.363% | |
BRF | −8.506 | 0.005 | 0.96 | 0.01 | 0.657% | |
AdaBoost | 0.0036 | 0.0045 | 0.74 | 0.034 | 1.941% | |
Validation | BPNN | −0.002 | 1.031 | 0.821 | 0.454 | 20.132% |
XGBoost | −0.0030 | 0.0049 | 0.84 | 0.028 | 1.312% | |
BRF | −0.0005 | 0.0033 | 0.78 | 0.027 | 1.382% | |
AdaBoost | −0.0037 | 0.0049 | 0.66 | 0.040 | 2.22% | |
Test | BPNN | −0.0002 | 0.005 | 0.821 | 0.030 | 1.547% |
XGBoost | 0.004 | −0.0002 | 0.831 | 0.025 | 1.288% | |
BRF | −0.001 | 0.004 | 0.851 | 0.026 | 1.284% | |
AdaBoost | 0.002 | 0.004 | 0.631 | 0.037 | 2.088% |
RF Optimization Method | R2 | MSE | MAE |
---|---|---|---|
BRF (Bayesian optimized random forest) | 0.851 | 1.125 | 1.045 |
TRRF (tuneRanger random forest) | 0.868 | 1.119 | 1.040 |
Study | Features Selected | Model | R2 |
---|---|---|---|
This Study | MNDCI(I), NDCI (443), Kd_490, BBP, NO3, O2, PO4, SPM, SSS, and Chl | TRRF | 0.87 |
Previous Study | Rrs490/Rrs555, Rrs490/Rrs510, Rrs510/Rrs555, Kd (490), CI, and Chl | RF | 0.73 |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Li, C.; Wu, H.; Yang, C.; Cui, L.; Ma, Z.; Wang, L. Advanced Machine Learning Models for Estimating the Distribution of Sea-Surface Particulate Organic Carbon (POC) Concentrations Using Satellite Remote Sensing Data: The Mediterranean as an Example. Sensors 2024, 24, 5669. https://doi.org/10.3390/s24175669
Li C, Wu H, Yang C, Cui L, Ma Z, Wang L. Advanced Machine Learning Models for Estimating the Distribution of Sea-Surface Particulate Organic Carbon (POC) Concentrations Using Satellite Remote Sensing Data: The Mediterranean as an Example. Sensors. 2024; 24(17):5669. https://doi.org/10.3390/s24175669
Chicago/Turabian StyleLi, Chao, Huisheng Wu, Chaojun Yang, Long Cui, Ziyue Ma, and Lejie Wang. 2024. "Advanced Machine Learning Models for Estimating the Distribution of Sea-Surface Particulate Organic Carbon (POC) Concentrations Using Satellite Remote Sensing Data: The Mediterranean as an Example" Sensors 24, no. 17: 5669. https://doi.org/10.3390/s24175669